Speaker: Prof.Ying Rong, Shanghai Jiao Tong University
Moderator: Prof. Liang Xu, School of Business Administration
Time: 14:00-15:00 , Apr. 15, 2021
Platform: Tencent Meeting 655 466 814
Organizers: School of Business Administration, Research Office
Dr. Ying Rong is a professor at Antai School of Economics and Management, Shanghai Jiao Tong University. He returned to China to teach at Shanghai Jiao Tong University in August 2010. Prior to that, he did postdoctoral research at the University of California at Berkeley and Lehigh University in the United States, where he received his bachelor's degree, master's degree and doctor's degree, respectively. His research interests include service operations management, retail operations management, operation of emerging business models, and data-driven optimization models. His papers have been published in international academic journals such as Management Science, Operations Research, Manufacturing & Service Operations Management, Production and Operations Management, Naval Research Logistics, and IIE Transactions. Professor Ying Rong is the recipient of the 2015 National Science Foundation for Distinguished Young Scholars and the 2020 National Science Foundation for Distinguished Young Scholars. He has won several international awards, including the two-time MSOM Award for Best Paper, the TSL Award for Best Paper and the INFORMS Energy, Natural Resources & Environment Young Researcher Prize.
We study stochastic periodic-review inventory systems with lost sales, where the decision maker has no access to the true demand distribution a priori and can only observe historical sales data (referred to as censored demand) and feature information about the demand. We propose two feature-based nonparametric inventory algorithms called the feature-based adaptive inventory algorithm and the dynamic shrinkage algorithm. Both algorithms are based on the stochastic gradient descent method. We measure the performance of the proposed algorithms through the average expected regret in finite periods: that is, the difference between the cost of our algorithms and that of a clairvoyant optimal policy with access to information, which is acting optimally. We show that the average expected cost incurred under both algorithms converges to the clairvoyant optimal cost at the rate of O(1/
). However, the feature-based adaptive inventory algorithm results in high volatility in the stochastic gradients, which hampers the initial performance of regret. The dynamic shrinkage algorithm uses a shrinkage parameter to adjust the gradients, which significantly improves the initial performance. The idea of dynamic shrinkage for the stochastic gradient descent method builds on a fundamental insight known as the bias-variance trade-off.